This chapter summarizes the discussions that took place at the end of the workshop. The focus of the discussions ranged from synthesizing key themes to identifying areas that need additional research and attempting to pinpoint the most promising approaches that meet SAMHSA’s goals for the measurement of trauma.
Evelyn Bromet (Stony Brook University) said that the field has matured in the past three decades and that the discussion of measuring trauma in children was especially interesting because researchers only began focusing on this topic in recent years. She noted that it has become clear from the presentations that trauma is ubiquitous, even if most people do not meet the criteria for PTSD, and that PTSD is more likely to be associated with some types of traumatic events than others. The interesting question is why the likelihood of developing PTSD varies so much by individual. She said that a typical limitation of existing studies is that they do not collect enough information about symptoms, beyond the PTSD symptoms spelled out in the Diagnostic and Statistical Manual of Mental Disorders (DSM). It may be that people who experience potentially traumatic events and do not have PTSD symptoms instead have other symptoms.
Another limitation of the studies to date, noted by Bromet, is that they do not ask about how people have dealt with a potentially traumatic experience in a positive way. Although there is increasing interest in the concept of resilience, there are few good resilience measures. Bromet said that collecting data on resilience would be one of the most important con-
tributions SAMHSA could make in the field. She also commented that in terms of survey design, asking everyone some basic questions and then selecting subgroups for detailed follow-up questions would be a useful way to focus the effort, because a lot is already known from prior research about which groups are at highest risk.
Bromet expressed concern about the limitations of interview protocols in general, because responses cannot be independently verified, and it is also not clear whether the respondent (or even the interviewer) really understands what each question is asking. She recalled a 1988 meeting on the definitions of trauma and its consequences, when using a trauma checklist was proposed. She said that she was concerned at the time about the lack of rapport building that is characteristic of the checklist method, and she was pleased to hear the discussion that reflected a better understanding of these issues today, in particular in the presentation by Benjamin Saunders (Medical University of South Carolina); see Chapter 4. Another way in which the field has changed is that there is a broader understanding of the types of traumatic events that may affect people, beyond personal traumas, which was highlighted in the presentation by Robert Ursano (Uniformed Services University of the Health Sciences); see Chapter 3. Bromet argued that the broader community traumas need to become integrated into the data collections, in order for the studies to remain relevant.
Bromet noted that one topic that was not discussed during the workshop is trauma in elderly people. There seems to have been little research on that topic, but perhaps focus groups could provide useful information on the experiences of elderly people and their families.
Hortensia Amaro (University of Southern California) said that new data collected by SAMHSA on the topic of trauma has the potential to inform policy and programs and really affect people’s lives. In that context, she noted that the United States is becoming increasingly more diverse, and data collections do not always pay adequate attention to that change. The same is true for special populations, such as nonhousehold populations, which are particularly important to consider in the context of trauma. The rates of trauma and negative consequences may be higher among some groups that are underrepresented in typical household surveys, such as homeless people and people in prisons. This likely under-representation can affect the estimates based on national surveys, as well as the programs and services available to those who are most marginalized in the society.
Amaro commented that a topic that has not come up is historical trauma. She argued that structural violence and chronic discrimination are significant issues in some communities. For example, there is literature that focuses on historical trauma in Native American populations, and
there is increasingly more scientific evidence about the negative effect of that trauma on health. She urged SAMHSA to keep these types of traumas at the center of attention.
Amaro also highlighted trauma in refugee and immigrant populations as issues that deserve more emphasis than they have had in the past. Many of the recent refugees currently in the United States have been through war and government persecution. There are also immigrants who are not refugees but come from countries where the likelihood of experiencing potentially traumatic events was high, due, for example, to the violence associated with drug trafficking in some of the Latin American countries. She also noted the trauma that may be experienced as a result of fear of deportation and separation from family. Amaro noted that there have been some recent studies focused on these populations, especially refugees, but measures are needed that can capture these experiences. Robert Pynoos (University of California, Los Angeles) agreed that the unique characteristics of some of the subpopulations have a large societal impact in terms of resources for services.
Dean Kilpatrick (Medical University of South Carolina) agreed with Amaro that many of the marginalized and understudied populations are likely to have a higher prevalence of exposure to potentially traumatic events. However, he pointed out that developing approaches to study these populations is challenging, and the design should be based on SAMHSA’s ultimate goals in this area. He noted that the U.S. Department of Housing and Urban Development does a focused survey of the homeless population every year and estimates the homeless population (those who are homeless on a given night) to be between 500,000 and 600,000. Considering the size of the U.S. population overall, including the homeless in a national survey would require a complex sampling frame. An added difficulty for some of the special populations, such as the prison population, is that there are a variety of rules and procedures about access for interviews.
Kilpatrick argued that for producing national estimates, a good household survey is what is needed, because that would capture most of the population. However, he added that SAMHSA also needs to devise a plan for collecting information about special populations, and for those data collections specialized surveys may be needed. Such specialized surveys could also gather relevant additional information, focused on the particular needs of those groups. The fact that this approach would be more difficult than a typical household survey need not be a deterrent.
Ursano commented that the active-duty military population is larger than the population of many states and, although the fact that they are not included in such surveys as the National Survey of Drug Use and Health (NSDUH) is not surprising, it is nonetheless an important consideration.
Jonaki Bose (SAMHSA) pointed out that the NSDUH captures reserve veterans and information on military families but agreed that the exclusion of active-duty military personnel is an important limitation for a survey on substance use and behavioral health.
In terms of SAMHSA’s primary goal of measuring PTSD using DSM-5 criteria, Kilpatrick said that the discussions and experience of previous surveys suggest that this is feasible to do in population-based surveys. There are suitable approaches for screening for exposure to potentially traumatic events. It is also possible to measure the most relevant related topics, with the understanding that this would further increase the length of the survey. He noted that the discussions also underscored the need for different strategies to measure trauma in children and adolescents, compared with the approaches used for adults. In the case of adolescents who are age 12 and older, it is possible to interview them directly about trauma-related topics, although the implications of mandatory reporting are important to consider.
Kilpatrick emphasized that problems of comorbidity are common, and PTSD is not the only outcome that needs to be considered in connection with exposure to potentially traumatic events. However, data on PTSD are especially useful from a clinical perspective because effective treatments exist. He also agreed with the speakers who argued that collecting additional contextual information about potentially traumatic events is useful and that asking about these topics is feasible in a survey. He said that in the case of multiple events, the information could be collected for at least some of the events, and, in particular, the most recent event. The context of the first traumatic event is also relevant because it is often something that happened in childhood during an important developmental period. If asking about the first event, the next step could be to ask whether there was another event that had either a bigger effect on the person or was worse. In other words, the questions could be introduced with some additional descriptive information, rather than just simply asking whether an event happened.
Terry Schell (RAND) offered a different perspective on the apparent tension between the goals to measure trauma in a nationally representative sample of the whole population and collecting sufficient data from subgroups. He argued that the goal should be to develop measures that work equally well in any population. He also noted that it is difficult to obtain funding for using a scale in a special population until it has been used in a nationally representative sample. Once a scale has been fielded with a national sample, the data from it can be used for comparisons with data from special population samples. If SAMHSA focuses on getting the measure right and collecting nationally representative data, it will enable others to more easily collect data with specialized samples.
Kathleen Merikangas (National Institute of Mental Health) pointed out that SAMHSA is interested in measuring behavioral health in several areas. Discussions at previous workshops, for example, about specific mental illness diagnoses with functional impairment highlighted the need to consider focusing on the individual rather than disorders, especially because of the widespread comorbidity.1 It may be that what is needed is to ask about whether people are currently impaired and suffering and then dig deeper to understand the causes or reasons. Many of the same people will have not only PTSD, but also a history of psychosis, substance abuse, and other difficulties. The effects of exposure to traumatic events can be exacerbated by other problems, such as anxiety. Merikangas argued that understanding these interactions may be more useful than simply counting the number of people with PTSD.
Larke Huang (SAMHSA) said that for SAMHSA it is useful to learn about how to translate clinical work to large-scale population surveys in any population group. She agreed with Merikangas that research that shows the pervasiveness of comorbidity highlights the need to understand how exposure to traumatic events contributes not only to PTSD, but also to other mental health issues, substance abuse, and risky behaviors. SAMHSA also wants to know why some people develop PTSD and other problems and other people do not, why some people develop more severe symptoms, and how resilience factors in. The agency also wants to better understand what steps can be taken to prevent negative outcomes, reduce their impact, and move people toward recovery. She said that the discussion pointed at several substudies that could be conducted and that perhaps a multistudy research agenda is needed to gain an in-depth understanding of some of the issues.
Huang noted that the discussion revealed that there are several methodological challenges specific to measuring exposure to traumatic events and PTSD. SAMHSA would need further input on whether a survey is the correct mechanism for measuring these topics, and if yes, how to implement such a survey. She pointed out that the Mental Health Surveillance Study collected data on several of the related issues that have been discussed at the workshop, such as mental health and homelessness, and the agency already has experience with collecting some of those data.
Benjamin Druss (Emory University) agreed with Merikangas that functioning seems to be the common element that ties together the dif-
1 National Academies of Sciences, Engineering, and Medicine. (2016). Measuring Specific Mental Illness Diagnoses with Functional Impairment: Workshop Summary. J.C. Rivard and K. Marton, Rapporteurs. Committee on National Statistics and Board on Behavioral, Cognitive, and Sensory Sciences, Division of Behavioral and Social Sciences and Education. Board on Health Sciences Policy, Institute of Medicine. Washington, DC: The National Academies Press.
ferent topic areas of interest for SAMHSA’s current effort of expanding data collections on behavioral health. It is clear that the interface of events and issues that occur in people’s brains and lives is complicated, but, ultimately, functioning is paramount: thus, functioning may be the goal, and researchers and analysts could work back from that goal in order to address what SAMHSA really needs to know about from a public health perspective.
Kilpatrick commented that PTSD as a construct has advanced the field because it communicates in a clear way that one can have the same symptoms and impairments as a result of a variety of different events. As discussed, PTSD in turn leads to a variety of health risk behaviors, such as substance abuse. In that sense, Kilpatrick argued, PTSD can be the unifying concept that helps researchers look at other related issues, especially health risk behaviors.
Merikangas noted the importance of understanding the course and sequence of events. She acknowledged that SAMHSA cannot collect data on the whole life span, but sometimes surveys that rely on the DSM to determine skip patterns do not ask a sufficient number of follow-up questions that could be relevant. People who are subthreshold are often skipped out of sections because they do not meet what is an arbitrary criterion. The concept of spectrum is useful, but survey instruments often need to arrive at a yes/no answer. Another challenge in the context of surveys, Merikangas said, is that sometimes people who are being treated are difficult to categorize without further clinical follow-up or additional understanding about their histories, and such information is beyond the scope of the types of data collections that SAMHSA intends to do.
Merikangas also discussed the importance of obtaining the perspectives of multiple respondents from a household. In the NSDUH, in most cases children over age 12 provide information for themselves, but there is also interaction with a parent to obtain permission. The parent who provides permission, or, ideally, both parents, could also be asked to complete some of these modules in order to obtain richer information and a family perspective.
Bose responded that such an approach would require changing the structure of the NSDUH, but that is not out of the question. She commented that the workshop raised a number of issues that cut across multiple domains, including special populations, measuring trauma in children, and traumatic brain injury. Some of the questions raised are related to feasibility. It is clear that SAMHSA will need to think about what its primary analytic goals are and perhaps discuss how those goals fit into broader goals of the U.S. Department of Health and Human Services. The workshop provided a good overview of the issues that can serve as a background for those discussions.
Nora Cate Schaeffer (University of Wisconsin) said that it appears that a comprehensive study would require a new data collection. However, it might be very useful to also have some questions on the NSDUH because that survey already collects other relevant data about drug and alcohol use. One question is whether additions to the NSDUH would result in information that expands what has already been done. She argued that perhaps SAMHSA should not think about these options as mutually exclusive.
Bose said that SAMHSA would want to learn more about what is possible to obtain from sets of questions that are brief enough to include in the NSDUH. SAMHSA would also like to better understand whether the short scales that might be available can be considered proxies for PTSD or for other outcomes of interest to the agency, such as other adverse health outcomes, increased substance use, or increased risk behaviors. Schaffer responded that to the extent one of the problems with the short scales is that they include variance that is due to such factors as depression or anxiety that originate from something other than a traumatic event, including these questions in a survey that already collects data on those other issues can help with beginning to partition the variance. She added that it appears that measuring both causes and effects in the current NSDUH would be a big challenge.
Graham Kalton (Westat) said that it appears that the primary goal is to measure PTSD, or other effects of traumatic events, and not exposures to traumatic events. Given that goal, it is not clear why it is not possible to ask about PTSD first and then follow up with questions to find out the cause, at least if the primary goal is to estimate the number of people with PTSD. He also asked the participants to weigh in on the role of clinical interviewers in the context of measuring PTSD. Based on the literature from other medical fields, it seems that lay interviewers with a standardized instrument can produce better quality data, and there are large cost implications associated with this decision.
Schell said that he generally prefers standardized interviews. Allowing interviewers to deviate from a standardized script can be especially counterproductive when the intent is to communicate a subtext to respondents to encourage them to provide information that they otherwise may not want to provide. Having looked at questions that interviewers came up with in these situations leads him to believe that this is not a good idea. He acknowledged that establishing rapport is important, but improvised questions lead to substantial variance, and they are less useful in a survey than in an evaluation for clinical treatment.
In response to Kalton’s suggestion to measure the symptoms associated with PTSD and then follow up with questions about a series of life
events to understand the relationship between them, Schell said that he agrees that building the definition of trauma exposure into the measurement of PTSD only complicates the problem. However, separating the two concepts is difficult, in part because the DSM integrates them. For example, re-experiencing is one of the criteria (see Chapter 2), and generally established instruments first define the range of events before asking about memories. He noted that the nonmilitary version of the PTSD Checklist has a very loose definition of a potentially traumatic event, and that approach seems to work. This solution would generally be more acceptable if the intent is a dimensional approach, rather than probable diagnoses. Kilpatrick noted that when the PTSD Checklist is used, there is typically a context already established, and he believes that it is always better to be as specific as possible in a survey and assure that the respondent understands exactly what is meant by the questions. He reminded participants that the quintessential characteristic of PTSD is that it is in reference to life experiences and cannot happen without those experiences.
Pynoos commented that in the DSM-5, the PTSD section is in a new chapter that is about the centrality of life experiences. There is a lot of research being conducted in the fields of medicine, public health, and psychology that is focused on the centrality of life experiences, of which traumatic exposures are one form. He used the example of his own research on Native American reservations where exposure to traumatic deaths happens at a rate that is much higher than in other communities. He argued that trauma is a cross-cutting issue and that SAMHSA has an obligation to go beyond mental conditions to understand the science of life experiences and its outcomes on health, on life, and on psychiatric and medical disorders. Druss commented that if the goal is to understand life experiences and individual context in order to gain a synthesized view that is similar to the social determinants of health, then perhaps the focus should not be on PTSD, which is an extreme and specific case.
Keane noted that he used to think that the adverse health outcomes associated with PTSD were secondary to risky behaviors, unhealthy behaviors, and addictions. In the Boston health care system of the U.S. Department of Veterans Affairs, there is a study under way involving around 500 participants who receive very comprehensive, full-day examinations and testing. The researchers are finding compromised cortical thickness among the recently returned war veterans with PTSD, who are often people in their late 30s, which is a major issue. More importantly, it seems possible that some of this might be mediated by the presence of metabolic syndrome, which does not appear to be related to the use of atypical antipsychotic drugs because, unlike with Vietnam veterans, atypical antipsychotic drugs are rarely used in this population.
Keane said that the group has also started publishing articles about the issue of accelerated aging, secondary to the exposures and the development of the conditions, suggesting that perhaps metabolic syndrome or telomere length or other similar factors are playing a very serious role in comorbidities that have previously been essentially considered byproducts of PTSD. It could also be that this accelerated aging is driven by a separate process, exacerbated by alcohol and drug use, but there could also be a unique ongoing neurological process that is secondary to these exposures. Keane said that given the prevalence of the problems, this new research indicates the possibility of a very serious public health problem and also helps provide a better picture of what may be most effective in terms of services provided to returning veterans and others with PTSD.
Kilpatrick asked Keane to clarify whether he believes that this supports the argument of measuring PTSD. Keane responded that he is conflicted about this issue. He said that surveys that cannot provide data that are conclusive are not useful. What would be needed are surveys that can provide convincing evidence to inform strategies for next steps. He added that this issue is not exclusively a mental health problem.
Pynoos clarified that when he emphasized the centrality of life experiences he was not suggesting that he would not include PTSD, because some of the symptoms of PTSD are very different from those of other mental health conditions. For example, reactivity to reminders is pervasive and can persist through a person’s life. This can have physiological responses, as shown in cardiovascular studies, and there is also a better understanding of the role of C-reactive protein. Keane commented that recovery from surgery can take several days longer in people with PTSD than in people who do not have PTSD, and there are many other medical consequences. Pynoos said that the latest developments in the field bring a new scientific perspective and that these discussions are at the forefront of that. It is not surprising that it is not always clear how to proceed.
Saunders pointed out that these psychological and neurological changes can have more severe consequences in children because they have many more life years ahead of them. From his perspective, this fact underscores the need to measure exposure to potentially traumatic events, rather than simply measuring PTSD symptoms, even if the latter were possible. There may be other processes related to the exposure that are not full PTSD and may not even be detectable.
James Jackson (University of Michigan) commented that while there are clearly some promising areas of emerging research, SAMHSA staff have been charged with a specific task. Kilpatrick responded that it is clear that it is possible to measure in a survey at least what the DSM-5 defines. In some ways, the task becomes easier than it was with using the DSM-IV because the DSM-5 definition no longer includes the “intense fear,
helplessness, or horror” criterion. It is also possible to measure a variety of life events, and it is clear what events are in the DSM-5. There is a lot of literature on this issue, and the challenge is to decide which approach to use. Jackson said that a lot of the research on exposure to potentially traumatic events, PTSD, and related outcomes is influenced by a period in time when it was assumed that traumatic events were rare. Measuring the events made sense in that context, but what was learned over time is that these issues are very common in the population, and the focus has become the etiology of the negative consequences of traumatic events. This development might require a readjustment of how researchers think about different approaches, particularly because truly understanding the nature of a traumatic event can be very expensive in a national survey.
Schell said that if one thinks of the range of events that can potentially lead to PTSD, the conclusion is that almost everyone has experienced at least one of them. Exposure to the types of events that are the most potent predictors of PTSD is much more limited. However, he agreed that adequately measuring the events would have to go beyond just asking the person whether they experienced any of the events on a list. He added that national surveys, such as the National Comorbidity Survey Replication, measured exposure to a broad range of traumatic events with follow-up questions to capture specific characteristics of the events. Work could be done to reduce the list of traumatic events to a smaller subset, which could then potentially be included in the NSDUH. For example, a list of the 10 most relevant events could be developed, if it can be demonstrated empirically that adding an 11th item would only contribute very small additional value. Schell said that research of this type has not yet been carried out. The challenge is that a broad range of events needs to be included in a nationally representative sample to understand what is important. Researchers have tried, for example, to compare data about rape victims to data about flood victims, from different samples, and they have found that what can be learned from these comparisons is more limited than what could be accomplished with national samples.
Schaeffer commented that it appears that while exposure to potentially traumatic events is common, the extreme stress responses are not as common. The relationship is moderated by a variety of variables, and it seems that if one wants to look at PTSD, collecting these additional data on intervening variables is essential in order to have a meaningful study.